Differences in resprouting ability are not related to seed size or seedling growth in four riparian woody species



    Corresponding author
    1. School of Marine and Tropical Biology, James Cook University, Townsville QLD 4811, Australia,
    2. School of Marine and Tropical Biology, James Cook University, Cairns QLD 4878, Australia and
      *Present address and correspondence: Caroline Chong, Australian Centre for Tropical Freshwater Research, James Cook University, Townsville, QLD 4811, Australia (tel. +61 74781 5215; fax +61 74781 5589; e-mail yfcaroline@gmail.com).
    Search for more papers by this author

    1. Australian Centre for Tropical Freshwater Research, Townsville QLD 4811, Australia
    Search for more papers by this author

    1. School of Marine and Tropical Biology, James Cook University, Townsville QLD 4811, Australia,
    Search for more papers by this author

*Present address and correspondence: Caroline Chong, Australian Centre for Tropical Freshwater Research, James Cook University, Townsville, QLD 4811, Australia (tel. +61 74781 5215; fax +61 74781 5589; e-mail yfcaroline@gmail.com).


  • 1Resprouting is a key plant attribute facilitating persistence in disturbance-prone environments. Resprouting ability in seedlings may depend on both developmental ontogeny and seed size. However, the relationships between these factors are not well explored, especially for woody species with comparatively small seeds and epigeal germination.
  • 2We investigated resprouting capacity in seedlings from four subtropical, riparian, Myrtaceous tree species, Melaleuca leucadendra, Asteromyrtus symphyocarpa, Eucalyptus camaldulensis var. obtusa and Tristaniopsis laurina, displaying these characteristics. We recorded resprouting in response to simulated disturbance as a function of seed mass and developmental age (5–150 days post-emergence) and examined the acquisition of resprouting ability in relation to growth and biomass allocation patterns.
  • 3Patterns of resprouting were distinct among species, but the acquisition of resprouting ability was not determined by seed mass. The ‘small’ seeded M. leucadendra and the ‘intermediate’ seeded E. camaldulensis showed unexpectedly high shoot resprouting vigour from cotyledon stage (70% resprouting at 5 days post-emergence), as well as greatest ongoing allocation to root mass and lateral root development. In contrast, in A. symphyocarpa (another species with ‘intermediate’ seed mass) and T. laurina (a ‘large’ seeded species) resprouting rates during early development were much lower (< 10%), although there was a trend towards increasing resprouting ability with age in A. symphyocarpa (> 150 days). Resprouting capacity was also independent of seedling size and relative growth rate.
  • 4Our results indicate that the size-dependency of resprouting capacity varies considerably among these species. This suggests physiological and morphological species traits other than those directly related to reserve size or relative growth rate may convey survivorship in river environments.
  • 5Our findings show that resprouting capacity was not related to seed size and seedling growth patterns in these four species. This is different to evidence from comparative studies undertaken in fire-prone and other temperate environments. A broader survey of seedling resprouting ability including more species is required to determine the generality of our findings in riparian species.


Resprouting is a critical trait for plants exposed to disturbances that cause the loss of most above-ground material (Bond & Midgley 2001; Del Tredici 2001; Cornelissen et al. 2003). Under such conditions, resprouting is competitively advantageous, because it provides the opportunity for new shoots to rapidly occupy opened areas (Whelan 1995; Vieira & Scariot 2006) as well as allowing faster return to reproduction (Whelan 1995; Peterson & Jones 1997; Bond & Midgley 2003).

In adult plants, resprouting capacity is often associated with allocation to root biomass, below-ground storage organs and/or larger non-structural carbohydrate (i.e. starch) reserves (Canadell & Lopez-Soria 1998; Landhausser & Lieffers 2002; but see Sakai & Sakai 1998). A similar relationship may also be expected for seedlings, because any selective advantage of carbon storage in adults should also be expected during the first season of plant development when resource demands for establishment are highest (e.g. Howe & Smallwood 1982; Schwilk & Ackerly 2005). Indeed, the amount of metabolizable reserves in a developing seedling is considered a key factor influencing the probability of establishment under generalized environmental hazards (Westoby et al. 1996; Muller et al. 2000; Walters & Reich 2000). If seedling reserve size promotes the ability to meet the physiological demands of development as well as stress tolerance (Saverimuttu & Westoby 1996; Boege & Marquis 2005), then species differences in resprouting capacity during deployment should reflect maternally provisioned reserves as a function of seed size and/or seedling morphology (Armstrong & Westoby 1993; Kitajima 1996; Green & Juniper 2004; but see Moles & Westoby 2006).

Nearly all the previous studies of resprouting ability in seedlings have focused on species that display hypogeal germination strategies. In these species, storage cotyledons remain in or on the soil, and disturbance generally removes the growing shoot without removing reserves (e.g. Harms & Dalling 1997). Because these species retain maternally provisioned resources, they also constitute the group in which resprouting might be expected (Kitajima 1996; Lehtilä 1999; Edwards & Gadek 2002). Only one study has examined resprouting ability at the earliest seedling stage in species with epigeal germination (Armstrong & Westoby 1993), even though these species may be subject to similar selective effects of disturbance as hypogeal species (e.g. Paz 2003). Understanding resprouting in seedlings from epigeal germinating species is very important, because, in contrast to hypogeal germinating species, resprouting capacity will rely on allocation strategies indirect of maternal provisioning (Lehtilä 1999; Stowe et al. 2000).

Physiological rates of activity (as well as stored reserves) may drive resprouting response, and vary due to changing functional priorities as a plant develops (Weiner 2004; Boege & Marquis 2005; Lamb & Cahill 2006). If resprouting in seedlings has similar functional value as it has in adult plants, then species differences in resprouting behaviour as seedlings might also be expected to reflect patterns of allocation to root and shoot growth associated with resprouting in adults (e.g. Schwilk & Ackerly 2005). In seedlings, allocation to root and shoot differs between species on the basis of seed mass (Westoby et al. 1992; Westoby et al. 1996), although seed mass might be secondarily related to resprouting, via the association between seed mass and relative growth rate (RGR, Wright & Westoby 2000; Aronne & De Micco 2004). Low RGR is associated with structural stability and below-ground resource storage (Westoby et al. 1992; Schwilk & Ackerly 2005), whereas rapid RGR is associated with fast turnover of above-ground plant parts supporting photosynthesis (Verdaguer & Ojeda 2002; Gurvich et al. 2005). Thus low RGR (associated with large seed size) might be expected to confer initial resprouting ability in seedlings.

Evidence for the ecological roles and evolutionary maintenance of resprouting in woody plants draws overwhelmingly from fire-prone and/or temperate ecosystems and adult plants (Bond & Midgley 2001; Pausas et al. 2004; Vesk & Westoby 2004; Vesk et al. 2004). Surprisingly few field or glasshouse studies have examined resprouting in systems where the main agent of disturbance is not fire, although resprouting as a mechanism for persistence should be expected in any environment prone to inescapable damage (e.g. Gom & Rood 1999; Paciorek et al. 2000; Spiller & Agrawal 2003).

Flood-prone river environments provide a unique framework for investigating resprouting. First, they represent high-stress areas where the main agent of disturbance is not fire. In subtropical dryland rivers in Australia, climatic and fluvial processes drive multiple agents of stress and disturbance, including high-force floods and submergence, seasonal aridity, variable soil resource availability, and sediment scouring and burial (e.g. Fielding et al. 1997; Puckridge et al. 1998; Pettit & Froend 2001). Secondly, the dominant tree species in these environments tend to be small-seeded epigeal germinators, ideal to examine relationships between growth traits and resprouting in seedlings that lack large cotyledonary reserves.

In this study, we use a glasshouse experiment with simulated disturbance to evaluate the interactions between maternal investment (seed size) and seedling development on resprouting capacity in four tree species from riparian areas in northern Australia. Specifically, we test whether differences in the acquisition and extent of resprouting ability reflect differences between species in these primary developmental attributes of seedlings.

Materials and methods

study species

We chose four plant species from high-discharge riparian ecosystems in northern Australia. All species selected were from the family Myrtaceae because this family is one of the most numerically dominant in these areas. Species were selected that (i) represented unique genera from distinct tribes within the Myrtaceae (see Wilson et al. 2005); (ii) primarily occurred in riparian habitats; (iii) had geographical distributions that were sympatric for part of their total range extents; (iv) were similar in growth form (trees); and (v) had a range of seed sizes across species that encompassed at least two orders of magnitude (Table 1). The four species were: Melaleuca leucadendra (L.) L., Asteromyrtus symphyocarpa (F. Muell.) Craven, Eucalyptus camaldulensis var. obtusa Blakely, and Tristaniopsis laurina (Sm.) Peter G. Wilson & J.T. Waterh. Resprouting has been previously documented in mature trees in the genera Melaleuca, Eucalyptus and Tristaniopsis, although there is no available evidence of resprouting behaviour in early seedlings of any of the species used here.

Table 1.  Summary description of species, distribution, habit, seed form and mean total seed mass (mg) for taxa used in this study; four riparian woody tree species from distinct Myrtaceous genera were used. Extra-Australian distribution localities: PNG = Papua New Guinea; IND = Indonesia. Australian localities: QLD = Queensland; NT = Northern Territory; WA = Western Australia; NSW = New South Wales; VIC = Victoria. Localities in italics correspond to localities of seed lots used
CodeSpeciesDistributionHabitatHabitSeed wingedSeed and cotyledon formSeed mass (mg) (% s.e.)
MELMelaleuca leucadendra sensu latoPNG, IND, QLD, NT, WAWatercourses; lagoonsTree to 30 mNoSeed narrowly obovoid/oblong; cot. obvolute0.07 (4.36)
ASTAsteromyrtus symphyocarpa (F. Muell.) CravenPNG, NT, QLDWatercourses; Melaleuca woodlands; monsoon rainforest; coastal sand plainsShrub or tree to 17 mYesSeed obovoid, wing well developed at seed apex; cot. plano-convex0.31 (3.05)
EUCEucalyptus camaldulensis var. obtusa BlakelyQLD, NT, WAWatercoursesTree to > 30 mNoSeed round/obovoid; cot. obovate0.41 (2.45)
TRISTTristaniopsis laurina (Sm.) Peter G. Wilson & J.T. Waterh.QLD, NSW, VICWatercourses, temperate rainforestTree to 30 mYesSeed contained in a samara; cot. obvolute1.60 (2.37)

Seeds from each species were obtained from commercial seed suppliers (Australian Tree Seed Centre, Australian Commonwealth Scientific and Industrial Research Organization, Canberra, ACT; Nindethana Seed Company, Albany, WA; Australia) and thus were collected from multiple parent populations. For M. leucadendra, we collected additional seed from reproductively mature individuals at Keelbottom Creek, Upper Burdekin River catchment, Queensland, Australia. Seeds of each species from different locations were pooled. We are confident all seeds represent the same ecological taxa because, in all cases, seed suppliers provided unambiguous taxonomic and geographical identifications, although we acknowledge intrageneric relationships in the Myrtaceae are subject to ongoing study (e.g. Craven & Lepschi 1999; Sytsma et al. 2004; Wilson et al. 2005).

seed size

Seed size was quantified as total dry seed mass. Characteristic of their genera, E. camaldulensis and M. leucadendra seeds were not winged, whereas T. laurina and A. symphyocarpa had winged structures that may not contribute directly to seed reserve mass (cf. Leishman & Westoby 1994). We report total seed masses for winged and non-winged species because subsequent analysis for a subset of seeds showed that removal of the winged structures had no effect on seed mass class assignations.

Seed mass was determined via random sampling of seeds drawn from four seed lots. For species with seed mass less than 1 mg (M. leucadendra, E. camaldulensis and A. symphyocarpa) an estimate of mean seed mass was calculated from 10 groups of 10 seeds. For T. laurina individual seeds were large enough that resolution of the balance allowed for mass determination in individual seeds. In this case we determined mean seed mass from a sample of 50 individual seeds.

growth conditions

Plants were grown in seedling tubes filled with a 6 : 3 : 2 steam sterilized sand/peat/perlite mix containing macronutrients (Yuruga Nursery Pty Ltd and Greening Australia Townsville, Queensland, Australia; Staypak super native tubes, c. 70 × 70 × 160 mm). The germination substrate was brought to moisture capacity before sowing. A capillary-driven watering system was used to maintain plants based on success in pilot growth trials (the ‘bog method’; Wrigley & Fagg 1993, pp. 18–21).

Plants were maintained in a glasshouse (School of Marine and Tropical Biology, James Cook University, Townsville, Australia). Periodic measurements of ambient temperature indicated a diurnal temperature range of c. 18–33 °C, conditions typical of regional records for the season (Bureau of Meteorology 2006). Measurements of incident light indicated ranges supportive of plant growth, although considerably lower than outside conditions (e.g. average photon flux density 403 µmol s−1 m−2, 60–80% shade; LI-COR photometer model LI-189).

experimental design: allocation to seedling tubes

Two hundred seedling tubes were randomly allocated to each of five PVC watering trays (each 1200 × 1200 × 150 mm). Within each watering tray, 10 blocks of 20 seedling tubes were placed in plastic seedling frames (in total, 1000 seedling tubes; 250 tubes per species). To overcome potential germination failure, 8–20 seeds were sown in each tube. Seeds were covered with a fine layer of steam-sterilized sand and moistened with a fine spray mist. We scored emergence as radicle protrusion to full expansion of both cotyledons from the seed coat. Where necessary, seedlings were subsequently thinned to two individuals per tube at 8–10 days post-emergence to minimize potential intraspecific competitive effects. Nevertheless, seedlings did not establish in every tube; thus replication was less than 250 in two species (A. symphyocarpa and T. laurina).

experimental design: allocation to clipping and biomass experiments

For each species, the 250 seedlings were randomly allocated to three groups: (i) clipping experiment (n = 120); (ii) biomass estimation experiment (n = 120); and (iii) control (n = 10). Within the ‘clipping’ and ‘biomass estimation’ groups, seedlings were further randomly allocated into one of six treatments (see below). Where we had a total of 120 seedlings available for each species in each experimental group, sample sizes for treatments within groups were 20 (M. leucadendra and E. camaldulensis). Because A. symphyocarpa and T. laurina were represented by only 200 and 224 seedlings, respectively, replication of individuals in each clipping treatment and in each harvest was 15 (A. symphyocarpa) and 17 (T. laurina).

experimental protocol: clipping

To examine resprouting ability with respect to seedling age we randomly assigned seedlings to six clipping treatments. The timing of the clipping treatments was 5, 10, 15, 25, 40 and 60 days following full expansion of the cotyledons. Clipping involved the removal of all stem and leaf material 5 mm above substrata level using surgical scissors. Resprouting and survival were monitored daily for 15 days post-clipping and every 2–3 days thereafter. For clipped individuals, we recorded initiation of resprouting response as the first visible formation of a shoot bud. For final assessment of resprouting response (new shoot formation vs. seedling death) we checked all paired seedlings for survival 3 months after clipping.

experimental protocol: biomass quantification

To examine the relationship between resprouting response and seedling growth patterns, we characterized plant biomass in randomly selected seedlings for each species at the time of each clipping treatment. Plants harvested for biomass quantification were independent of those assigned to clipping treatments.

We recorded leaf developmental stage (cotyledon persistence and number of expanded foliar leaves), fresh shoot lengths (hypocotyl and epicotyl), fresh total root length, and numbers of lateral roots (a) arising from the root base (‘basal lateral roots’) and (b) in total. Individual plants were extracted from the sandy substrate and separated into cotyledon/leaf, shoot and root parts. All soil was removed with the aid of a fine brush under running water. Dry masses (65 °C, 48 h) of roots, shoot and leaves of each plant were measured. Root, shoot and leaf mass fractions (RMF, SMF, LMF, respectively) were calculated as component mass/total plant dry mass (Wilsey & Polley 2006). Relative growth rates (RGR; mg mg−1 day−1) for the period 5–25 days after emergence were estimated as the slopes of Model I regression of ln total plant dry mass against days elapsed.

data analyses

Differences in emergence rate among species

Species differences in mean emergence time were evaluated using anova. We utilized the Monte Carlo Markov Chain (MCMC) algorithm and uninformative priors in WinBUGS v.2.10 (Spiegelhalter et al. 2005), incorporating M. leucadendra as the reference class. A normal distribution was assumed for the species fixed effect. Within-species variances were modelled as belonging to gamma distributions with r = µ = 0.001. Preliminary model selection analyses using the deviance information criterion (DIC) indicated this single-factor model to be most parsimonious. We generated 100 000 samples from the model posterior distribution, discarding the initial 10 000 samples as a burn-in, for each of three independent Markov chains (thus, total sample size was 300 000 samples from the three chains). The mean and standard deviation of the model coefficients, and the 2.5th and 97.5th percentiles of the posterior distribution (i.e. the Bayesian 95% credible interval; Johnson 1999; Parris 2006) were calculated.

Differences in resprouting ability among species

Bayesian logistic modelling using uninformative priors was used to identify the effects of species, seedling age (as a surrogate for developmental stage) and the interaction between these factors on resprouting ability (i.e. proportion of seedlings that survive clipping). To evaluate the a priori hypothesis that seedling resprouting ability is a function of seed size, species’ resprouting capacity was investigated relative to the reference class M. leucadendra, the smallest-seeded species. We considered this modelling approach appropriate because a binomial response was predicted in the data, i.e. individuals either resprouted or died. The continuous explanatory variable (age) was standardized (standardized value = (xage – mean xage)/SDage)) to improve sampling efficiency and help reduce autocorrelation between successive MCMC samples. Preliminary model selection analyses using the DIC and parameter posterior probabilities identified this model to best fit the data, compared with an alternative parameterization accounting for a species random effect. Hence, for this data set, total variance associated with regression of resprouting ability on age appears to be estimated validly by variation due to differences in probability with age among species. Again, we used 300 000 samples from the model posterior distribution (100 000 samples from each of three independent Markov chains, discarding the initial 10 000 samples as a burn-in) to estimate model parameters.

Differences in growth patterns among species

We began by performing discriminant function analysis (DFA) to evaluate species differences in growth characteristics in a multivariate context (JMPIN v.4.0.3, SAS Institute Inc.). DFA accounts for group structuring by generating linear combinations of variables that maximize between-group to within-group variance (Quinn & Keough 2004). We evaluated nine growth variables via DFA: RMF, SMF and LMF; number of leaves; total shoot length (TSL); total and specific root length (TRL, SRL); and number and frequency of lateral roots.

A key interest in this study was to determine whether resprouting ability among species differed on the basis of differences in seed size and seedling growth variables. Because allocation to root biomass was a strong indicator of species growth trends in preliminary DCA, we investigated relationships between root investment variables and resprouting among species via standardized major axis analyses ((S)MATR; Wright et al. 2001; Falster et al. 2003; Warton et al. 2006).

Relationships between resprouting ability, seed size and growth patterns among species

To examine relationships between growth parameters and resprouting ability on the bases of species and seed mass, we compared species identity (ranked in order of increasing seed mass) with species growth traits (ranked in order of increasing magnitude), for the following parameters: RGR, specific root length, lateral root investment, total root biomass and resprouting ability.

Assessment of growth rate, biomass ratios and allometric trends focused on data for seedling growth up to 25 days. This was because growth tube length became limiting for some individual seedlings after 25 days, potentially altering growth/allocation patterns.

Data were natural log or arcsine (ratios) transformed where appropriate prior to all analyses.


differences in seed mass among species

Species mean seed masses spanned two orders of magnitude. Mean M. leucadendra seed size (0.07 mg) was less than 25% that of A. symphyocarpa and E. camaldulensis, and less than 5% that of T. laurina (Table 1). We ranked species based on mean seed mass as ‘small’ (M. leucadendra), ‘intermediate’ (A. symphyocarpa and E. camaldulensis) and ‘large’ (T. laurina).

differences in emergence rate among species

Germination was rapid across all species (5–8 days). Time to emergence was greater among T. laurina and A. symphyocarpa seedlings compared with E. camaldulensis and M. leucadendra, although overall, differences in time to emergence among species were less than 5 days (i.e. relative to M. leucadendra, E. camaldulensis mean emergence time +0.05 days, 95% CI –0.06–0.16; A. symphyocarpa, +2.54 days, 95% CI 2.44–2.64; T. laurina, +4.09 days, 95% CI 3.99–4.19; Table 2). Within a species, variability in time to emergence was lowest in M. leucadendra and highest in A. symphyocarpa (Table 2).

Table 2.  Coefficients (mean, standard deviation, and 2.5th and 97.5th percentiles) of the explanatory variables (tspec = time in days to emergence from radicle protrusion to full cotyledon expansion within species; σ2 = variance within species) included in the best-fit anova model, for mean time from germination to full emergence among species. Values of species coefficients represent terms relative to the reference class tspec = MEL. Refer to Table 1 for species codes
inline image2.9300.2682.4283.478
inline image4.8480.5123.8975.901
inline image2.5420.2322.1063.017
inline image4.2800.4243.4905.149

differences in growth traits among species

As anticipated, total plant mass at 5 days post-emergence was tightly correlated with seed mass (y = 0.116 + 0.95x, r2 > 0.99, P = 0.0007). At this developmental stage, epigeal cotyledons constituted total leaf mass and represented a large component of total mass allocation in all seedlings (cotyledon mass fraction 0.58, 0.75, 0.63 and 0.68 in M. leucadendra, A. symphyocarpa, E. camaldulensis and T. laurina, respectively; Table 3). Shoot architecture varied among species at the cotyledon stage (specific hypocotyl length 157.10, 92.32, 128.00, and 37.66 mm mg−1 dry shoot mass in M. leucadendra, A. symphyocarpa, E. camaldulensis and T. laurina, respectively; Table 3). By 25 days seedling development, the largest-seeded T. laurina showed proportionately lowest total mass accrual. Relative growth rate was distinct between all four species (RGR 5–25 days: T. laurina < M. leucadendra < A. symphyocarpa < E. camaldulensis; Table 4).

Table 3.  Summary of seedling growth characteristics for four Myrtaceous riparian woody tree species grown under glasshouse conditions. Data values refer to cotyledon stage (5 days post-emergence) or seedling stage (25 days post-emergence) as indicated. Values represent means of back-transformed data from arcsin square root (cotyledon mass fraction, root mass fraction) and ln(specific hypocotyl length, specific root length) transformations; other values as indicated. Refer to Table 1 for species codes
Cotyledon stage
Cotyledon mass fraction at 5 days (proportion total dry plant mass)  0.58 0.75  0.63 0.68
Specific hypocotyl length at 5 days (mm mg−1 dry shoot mass)157.1092.32128.0037.66
Seedling stage
Root mass fraction(proportion total dry plant mass)  0.21 0.11  0.18 0.17
Specific root length (mm mg−1 dry root mass)146.2067.10 35.2061.30
Lateral root frequency (mean no. laterals mm−1 root)  0.58 0.33  0.26 0.23
Lateral root investment (mean no. laterals mg−1 dry root mass) 84.8021.90  9.3014.70
Total root length; mean (SE) 28.00 (1.65)69.10 (9.17)226.70 (9.33)76.40 (6.89)
Number of lateral roots; median (range) 15 (10–22)21 (11–24) 59 (32–103)19 (3–43)
Number of basal lateral roots; median (range)  2 (0–6) 1 (0–3)  0 (0–3) 0 (0)
Table 4.  Comparison of species relative growth rate (RGR) calculated from ln(total plant mass, mg) and ln(time post-emergence, days) for t = 5–25 days post-emergence. Values are results of SMA analyses for individuals of each species. Lower and upper β represent 95% confidence intervals for the estimated SMA slopes. There were significant differences among all species slopes (SMA common slope test, P = 0.001; pair-wise comparisons, P < 0.0001). Also shown are Model 1 slopes (RGR, mg mg−1 d−1) for comparative purposes. Refer to Table 1 for species codes
Speciesnr2PSMA slopeLower βUpper βModel I slope
MEL350.84< 0.00011.471.271.691.34
EUC650.92< 0.00013.132.913.373.00
AST500.90< 0.00012.151.962.352.03
TRIST680.83< 0.00010.980.891.090.90

Compared at 25 days post-emergence, shoot height differed among species (mean total shoot length 8.66 mm, 23.77 mm, 61.29 mm and 21.59 mm in M. leucadendra, A. symphyocarpa, E. camaldulensis and T. laurina, respectively), whereas allometric trends in leaf development were similar (all seedlings three to four true leaves; data not shown).

Discriminant function analysis indicated strong separation of species measured at 25 days (first discriminant function, eigenvalue = 40.98, explained variance = 98.8%; manova, Pillai trace = 1.877, F27,186 = 11.514, P < 0.0001). In large part, the separation was due to differences between species on the basis of below-ground plant parts, and relative allocation to root biomass was identified as a strong indicator of growth trends among species. For example, root mass fraction, lateral root frequency, lateral root investment and specific root length contributed most to discriminant functions 1 and 2, whereas number of leaves and shoot and leaf mass fractions contributed least. Because of this we only report bivariate relationships between root biomass variables across species via SMA analyses (below). Growth tray (i.e. blocked position of plant in glasshouse) had no effect on species development (multi-response permutation procedure, A = −0.024, P = 0.953).

differences in root characteristics among species

At 25 days post-emergence, seedlings in all species allocated resources to root mass relative to total plant mass at a similar rate ((S)MATR statistic = 1.491, P = 0.703; common slope estimate = 1.18, 95% CI 0.97–1.45). Among species, mass accrual was lowest in A. symphyocarpa and M. leucadendra and highest in E. camaldulensis. Significant shifts in elevation in both the 5–25 day and 25 day data sets (post hoc pair-wise comparisons, P < 0.05) indicated that M. leucadendra showed greatest investment in root mass when compared at the same absolute plant mass (all pair-wise contrasts, P < 0.0001) (Fig. 1).

Figure 1.

The relationship between log total plant mass and log root mass among seedlings within four tree species from subtropical, riparian northern Australia. Values represent the results of SMA analyses across individual plants at 25 days post-emergence. Slopes were homogeneous across species ((P = 0.703), common slope of 1.18 (0.97, 1.45)). There was a significant elevation shift between M. leucadendra and other species: log root mass at a given log total plant mass was greatest for M. leucadendra (see text for details). Lines represent the linear regression of individual values for M. leucadendra (solid line) and for the three other species combined (dashed line). Closed circle =M. leucadendra, open triangle = A. symphyocarpa, open square =E. camaldulensis var. obtusa, open diamond = T. laurina.

differences in resprouting among species

Despite the assumed severity of the clipping treatment, seedlings of M. leucadendra and E. camaldulensis showed high resprouting success, even at very early seedling ages. For example, at 5 days post-emergence, recovery from clipping was 70% in both species, and we observed initiation of new shoots from multiple stem epicormic buds, typically 7–12 days after clipping. In contrast, there was minimal resprouting in T. laurina and A. symphyocarpa in 5–60 day old plants (in total, 3 of 102 and 3 of 90 individuals of each species resprouted, respectively). The failure of clipped seedlings to survive was not due to factors other than the clipping treatment because the unclipped paired controls in all four species experienced 0% mortality across the experiment.

Logistic modelling provided strong support for an important effect of species on resprouting ability. In particular, resprouting ability over 150 days growth was predicted to be substantially lower in A. symphyocarpa and T. laurina than in M. leucadendra (A. symphyocarpa, mean –4.49, 95% CI –5.82 to –3.40; T. laurina, mean –7.13, 95% CI –11.67 to –4.59; Table 5). In comparison, evidence in support of differential resprouting ability in E. camaldulensis relative to M. leucadendra was less strong, and positive (E. camaldulensis, mean 1.64, 95% CI 0.49–3.09; Table 5), but at 25 days resprouting ability in M. leucadendra and E. camaldulensis was similar. Evidence in support of the effect of age alone on resprouting ability was equivocal, with the 95% credible interval encompassing 1 (Table 5), whereas the species–age interaction had positive effects on resprouting ability in A. symphyocarpa and E. camaldulensis relative to M. leucadendra (A. symphyocarpa, mean 1.68, 95% CI 0.97–2.42; E. camaldulensis, mean 2.74, 95% CI 0.99–4.90).

Table 5.  Coefficients (mean, standard deviation, and 2.5th and 97.5th percentiles) of the constant and explanatory variables (specX = species x, age = time to emergence, interaction) included in the best-fit logistic regression model, for seedling resprouting response to a single clipping treatment at seven age harvests (5, 10, 15, 25, 40, 60, 150 days post-emergence). Clipping treatment comprised removal of all plant material 5 mm above substrata. Response to clip was binary; plants either resprouted or died. All observed resprouts were from stem epicormic regions. Values of species and interaction coefficients (specX, spec × ageX) represent terms relative to the reference class spec = MEL. Refer to Table 1 for species codes
spec * ageAST1.6800.3700.9752.427
spec * ageEUC2.7420.9990.9894.900
spec * ageTRIST−3.5972.601−9.8490.011

relationships between resprouting ability, seed size and growth traits among species

Species’ relative resprouting capacity was consistent with general species trends in root mass investment (i.e. root mass fraction). The two species demonstrating high resprouting ability throughout early growth (M. leucadendra and E. camaldulensis) also showed intermediate to highest investment in root biomass relative to absolute plant mass, although we found no evidence that this investment was associated with development of a lignotuber at 25 days growth. There was no evidence for relationships between resprouting capacity and species ranked in order of seed mass, relative growth rate or absolute seedling size (Table 6).

Table 6.  Summary of relationships between seedling resprouting ability and growth parameters among species at 25 days post-emergence. For each parameter, species codes are listed in order of increasing magnitude of the measured parameter. Superscripts represent species groups that are significantly different (for resprouting ability and RGR, refer to statistical analyses in Table 5 and Table 4; for seed mass, total biomass and root mass fraction, groups represent results of SMA analyses: all pair-wise contrasts, P < 0.01). Refer to Table 1 for species codes
OrderResprouting abilitySeed massRGRTotal biomassRoot mass fraction


relationships between riparian seedling resprouting ability and reserve size

Differences between species in seedling resprouting ability have previously been correlated to larger seed and seedling size. This suggests that the amount of stored reserves is a principal mechanism promoting survival during early plant development (Westoby et al. 1992; Garwood 1996; Kitajima & Fenner 2000). If resprouting was related solely to seed mass, then it should be most prevalent and acquired at earlier stages of development in seedlings from species with greatest seed mass. This was not the case in our study. Rather, we found that resprouting was acquired earliest in seedlings of M. leucadendra and E. camaldulensis, and these species had small and intermediate seed masses. Moreover, survival rates in seedlings of these species were high even after clipping at 5 days post-emergence. At this stage, only epigeal cotyledons had developed, and plant functioning was most probably still reliant on seed-derived reserves (e.g. Kitajima 1996; Boege & Marquis 2005). Seedlings from A. symphyocarpa, however, showed minimal survival under clipping. This was also the case in seedlings from the larger-seeded T. laurina, despite mean seed mass in this species being more than three times that of E. camaldulensis, and 20 times that of M. leucadendra.

Furthermore, survival and resprouting were not related to absolute seedling size. Resprouting was lowest in T. laurina, although absolute seedling size in this species was substantially larger than in M. leucadendra and E. camaldulensis, both strong resprouters. One potential explanation is that species differences in resprouting at deployment stage may arise due to differences in allocation to metabolic activity in stem meristem regions (Boege & Marquis 2005). The finding of significant epicormic resprouting in M. leucadendra and E. camaldulensis indicates cotyledonary bud primordia are abundant and activated early in these species (cf. Pascual et al. 2002; Verdaguer & Ojeda 2002), which, in turn, may enable seedlings to rapidly recover photosynthetic function even at very small plant body sizes.

Furthermore, our results suggest that considerable variation exists in species’ programmes of functional development from cotyledon stage that may not be simply determined by RGR, or the interplay between seed size and RGR (Vuorisalo & Mutikainen 1999; Muller et al. 2000; Boege & Marquis 2005). We found no evidence that resprouting during seedling development was a simple function of RGR, or plant size at seed, seedling or mature life stages, and thus no support for the predicted ‘cost’ of resprouting ability to growth potential that may arise due to the inverse relationship between growth rate and size of stored reserves (e.g. Pate et al. 1990; Gurvich et al. 2005; Pratt et al. 2005; but see Schwilk & Ackerly 2005). Although M. leucadendra and E. camaldulensis (both strong resprouters) showed intermediate to highest root allocation, we found no evidence that this was due to investment associated with development of a lignotuber during early growth. One potential explanation for the finding that resprouting capacity was not related to RGR or seedling reserve size might be that species differ in their rates of physiological efficiency (cf. Walters & Reich 2000; van Eck et al. 2004). For example, seedlings of M. leucadendra showed high shoot resprouting capacity in addition to greatest ongoing allocation to root biomass, indicating metabolic activity in both shoot and root modular regions is high (e.g. Vuorisalo & Mutikainen 1999). Efficient resource assimilation and translocation could have functional value during early development (e.g. Garwood 1996), because physiological efficiency can provide a mechanism allowing resprouting and recovery from damage and stress (Mulligan & Patrick 1985; Sakai & Sakai 1998; Reekie 1999; Iwasa 2000) despite small absolute biomass or reserve size. Explicit tests of this proposal are required. For instance, the relative contribution of resource assimilation rate to species differences in seedling functioning could be examined by comparing species’ growth and resprouting patterns as functions of photosynthetic rate per unit cotyledon and leaf nitrogen (Kitajima 1996; Wright & Westoby 2000). Nevertheless, our results implicate more than one solution to the requirements of early acquired resprouting ability. Seedlings of M. leucadendra and E. camaldulensis showed distinct allocation patterns, despite having similar resprouting capacities.

the roles of resprouting as a functional plant attribute

Our results also indicate that species differ in the acquisition of resprouting capacity during early seedling development. In support of this contention, an ad hoc clipping treatment at 150 days (four leaf pairs) revealed resprouting ability in A. symphyocarpa (12 of 15 individuals) but not in T. laurina (0 of 17 individuals), whereas the responses of M. leucadendra and E. camaldulensis were maintained across developmental stages. Thus, seedling resprouting ability is not well predicted by either plant size or age in M. leucadendra and E. camaldulensis (cf. Muller et al. 2000; Weiner 2004), but appears cued by increasing developmental stage in A. symphyocarpa. To some extent this must also be true for T. laurina, because resprouting has been shown in saplings and adult plants (e.g. Melick 1990a). However, it was not developed at 150 days post-emergence in our study. This is consistent with resprouting response being a widely variable phenomenon among species within disturbance type, growth form and size-stages (Vieira & Scariot 2006).

Plants that exist in environments that impose inescapable, recurrent disturbances causing large biomass loss should experience strong selective pressure for maintaining resprouting ability throughout their lifetime (Vieira & Scariot 2006). Resprouting has been shown in many mature-stage Myrtaceae, including members of all four genera studied here (e.g. Conde et al. 1981; Lacey & Johnston 1990; Melick 1990a,b; Miwa et al. 2001; Tierney 2004). By demonstrating that resprouting ability is acquired by cotyledonary stage in two ecologically conspicuous riparian epigeal taxa (M. leucadendra and E. camaldulensis) and, to a lesser extent, within the first year in another two riparian species (A. symphyocarpa and T. laurina), we provide support for the general proposition that resprouting is a generalized response to disturbance at the seedling stage as well as in adult plants (Bellingham & Sparrow 2000; Bond & Midgley 2001, 2003; Vesk & Westoby 2004).

An essential complementary consideration is that processes conferring resprouting ability may be advantageous for alternate functions, including nutrient and space acquisition (Jackson et al. 1999), stress tolerance and structural stability (Melick 1990a; Mensforth & Walker 1996; Davies & Giblin-Davis 2004; Pratt et al. 2005) and growth (Fielding et al. 1997; Gurvich et al. 2005), all of which may exert strong selective influence. Indeed, early seedling establishment will depend on the contributions of multiple plant parts (e.g. seed reserves, cotyledons and leaves as well as roots and shoots) to multiple functions (photosynthesis and reserve storage as well as nutrient uptake, anchorage and stability; Garwood 1996). Arguably, it is unlikely that disturbance acts as a sole selective pressure governing resprouting ability. More plausibly, resprouting reflects a suite of characteristics that confer benefit in the face of numerous environmental constraints and hazards.


This study demonstrates that M. leucadendra and E. camaldulensis show high resprouting ability from early seedling developmental stages, which is not a function of resource provisioning via large seed size, nor relative growth rate. Our results also indicate considerable between-species variation in seedling resprouting capacity, suggesting physiological and morphological traits other than those related to reserve size or relative growth rate alone may also contribute to resprouting and survivorship in river environments (Malanson 1993; Fielding & Alexander 2001). High allocation to meristematic activity might best explain resprouting, which may be selected on the basis of advantages it conveys for other life-history requirements such as establishment, structural stability and resource acquisition (e.g. Melick 1990a; Noland et al. 1997; Jackson et al. 1999; Abernethy & Rutherfurd 2001). This may explain why we were unable to demonstrate an association between resprouting ability and either seed size or RGR predicted from previous studies from fire-prone temperate areas. Studies of seedling resprouting ability that include larger numbers of species from riparian systems will be required before generalizations should be attempted.


We warmly thank Richard Pearson and Michelle Ensbey for useful discussion of statistical analyses, and Allyson Lankester, Chris Gardiner and Ainsley Calladine for logistic support. Peter Bellingham and two anonymous reviewers provided very helpful advice that greatly improved the manuscript. Support was received from the Australian Centre for Tropical Freshwater Research, the Cooperative Research Centre for Tropical Savannas Management and an Australian Postgraduate Award to CC.